Search

CN-121980814-A - Asphalt pavement rut local accumulation gray prediction method based on sine drive periodicity

CN121980814ACN 121980814 ACN121980814 ACN 121980814ACN-121980814-A

Abstract

The invention relates to the technical field of bituminous pavement track depth prediction, in particular to a bituminous pavement track local accumulation gray prediction method based on sine drive periodicity, which comprises the steps of 1, defining a sequence, 2, introducing sine drive items and period correction items, 3, constructing seasonal weakening buffer operators SAWBO, 4, optimizing parameters in pretreatment by adopting a Drosophila optimization algorithm FOA, 5, constructing a dynamic local accumulation generation sequence, and 6, solving the number of local accumulation Step 7, constructing SAWBO-DPDGSM (1, 1), estimating parameter vectors, and step 8, and applying cases. The method is superior to the traditional statistical model and the exponential smoothing method, has obvious precision advantage and stability advantage in comparison with a machine learning model, and is more suitable for track depth sequence prediction with obvious trend evolution and seasonal fluctuation characteristics.

Inventors

  • XU XUNQIAN
  • ZHANG JUNHAO
  • JI HAIPING
  • ZHANG CHEN
  • JIN CHUNHUA
  • ZHOU ZIJUN
  • YE DANLI
  • ZHU JIAHUI
  • WANG YIWEN

Assignees

  • 南通大学

Dates

Publication Date
20260505
Application Date
20260210

Claims (9)

  1. 1. The method for predicting the local accumulated gray of the ruts of the asphalt pavement based on the sine driving periodicity is characterized by comprising the following steps of: step 1, data acquisition and sequence definition, namely acquiring a target road section rut depth moon detection sequence, and defining an original sequence as ; Step2, introducing a sine driving term and a period correction term to preprocess the data, namely preprocessing the original data Periodic processing, introducing sinusoidal driving terms And period correction term Obtaining a new sequence ; Step 3, constructing a seasonal weakening buffer operator SAWBO, wherein the seasonal period is set as Will be Seasonal location Divided into a plurality of sub-sequences and define the first The seasonal average value of each seasonal position is For the treated season average value Constructing a seasonal weakening buffer generating sequence ; Step 4, solving parameters in the pretreatment process, namely adopting a Drosophila optimization algorithm FOA to solve the parameters in the pretreatment process 、 And Searching to minimize the error of the fitting segment; Step 5, constructing a dynamic local accumulation discrete gray seasonal model DPDGSM (1, 1) generation sequence, namely generating a sequence for one-time seasonal weakening buffering Performing dynamic local accumulation once to obtain a dynamic local accumulation generation sequence ; Step 6, solving the local accumulation number Traversing all of the domain by grid search The value, the optimal local accumulation number is obtained by taking the test set prediction result MAPE as the minimum principle; step 7, constructing a sinusoidal driving periodic dynamic local accumulation discrete gray season model SAWBO-DPDGSM (1, 1) model and estimating parameters based on the obtained sequence Constructing a sinusoidal driving periodic dynamic local accumulation discrete gray season model SAWBO-DPDGSM (1, 1) model, and estimating a parameter vector by using a least square method; And 8, carrying out case application, namely predicting the track depth of the asphalt pavement.
  2. 2. The method for predicting the local accumulation gray of the rutting on the asphalt pavement based on the sinusoidal driving periodicity as set forth in claim 1, wherein in the step 1, the method comprises the following steps: Acquiring a target road section rut depth moon detection sequence, and defining an original sequence as : (1) Wherein, the Is the first Observation of the rut depth in the period, the sample length is 。
  3. 3. The method for predicting the local accumulated gray of a rutting on an asphalt pavement based on sinusoidal driving periodicity as recited in claim 1, wherein in the step 2, the method comprises the steps of: For the original sequence Seasonal processing to eliminate the effects of seasonal variations, introducing sinusoidal drive terms For characterizing seasonal influence of independent variables on dependent variables, periodic correction terms Then the original data is used to reflect the periodic fluctuations of the dependent variable After pretreatment, a new sequence is obtained ; (2) (3) (4) Wherein, the Representing the amplitude of the sinusoidal drive term, Is the frequency of the cycle time, Is the phase of the light and the phase of the light, 、 And Is the coefficient of the periodic correction term.
  4. 4. The method for predicting partial accumulated gray of bituminous pavement rut based on sinusoidal driving periodicity as set forth in claim 1, wherein step 3 comprises the steps of setting the seasonal period as Will be Seasonal location Divide into a plurality of sub-sequences, let the first The number of the data corresponding to the positions of each season is : (5) Wherein: Is the first The number of data corresponding to the positions of the seasons, ; And define the first The seasonal average value of each seasonal position is : (6) Wherein, the Moon data fetch Quarterly data retrieval , In order to calculate the sum of the variables, ; To maintain periodic structure while weakening early disturbance, a seasonal average weakening buffer generating sequence is constructed : (7) (8) Wherein, the Is the first Seasonal average of individual seasonal positions, called Is a pretreated sequence A sequence of seasonal weakening buffer generation, the corresponding operation is called a seasonal weakening buffer operator (SAWBO), wherein, Is a period of the season, Is the length of the sample and, The k-th element in the time sequence, Is the first A seasonal average of the locations of the individual seasons, , Is the first The number of data corresponding to the positions of the seasons.
  5. 5. The method for predicting the local accumulated gray of the rutting on the asphalt pavement based on the sinusoidal driving periodicity according to claim 1, wherein the step 4 comprises the following steps of establishing the following objective functions and constraint conditions for searching the optimal parameters: (9) Wherein, the And Is a weight coefficient for balancing the effects of MAPE and RMSE, Is a punishment factor for controlling punishment strength of constraint conditions; is a penalty term calculated from the constraint, and if a certain parameter is out of a predetermined range, a corresponding penalty is added.
  6. 6. The method for predicting the local accumulation gray of the rut on the asphalt pavement based on the sine drive periodicity of claim 1, wherein in the step 5, the method comprises the steps of introducing a dynamic local accumulation generating operator, setting the local accumulation order as Defining the dynamic local accumulation generation sequence as : (10) Wherein the method comprises the steps of Is the number of elements in the original sequence, Is the number of the partial accumulation, Representing sequences Is the first of (2) The number of elements to be added to the composition, 。
  7. 7. The method for predicting partial accumulated gray of bituminous pavement rutting based on sinusoidal driving periodicity according to claim 1, wherein step 6 comprises the steps of Solving and accumulating the number The new model is predicted in two stages from the data self-adaption angle; representing the number of original sequences, Representing the prediction step length, will be 1 st to 1 st The data is used as training set of the first stage, the first stage To the point of The data is used as a test set of the first stage, and all the defined domains are traversed by a grid search method The value of the sum of the values, Define a domain as Acquiring the optimal local accumulation number by taking the minimum of the test set prediction result MAPE as a principle; Will be the first To the point of The data is used as a training set of a second stage, the first stage To the point of The data is used as a test set of the second stage for constructing a model and performing effect analysis, and is determined according to the first stage The value is used for constructing a dynamic local accumulation generation sequence, and a least square method is used for obtaining a parameter vector of a model as And further obtaining fitting and predicted values of the new model, and evaluating modeling effects of the model according to error criteria.
  8. 8. The method for predicting partial accumulated gray of bituminous pavement rut based on sinusoidal driving periodicity according to claim 1, wherein step 7 comprises the steps of Performing one-time local accumulation generation, namely: (11) (12) the formula (11) is SAWBO-DPDGSM (1, 1) model, wherein Representing the number of cycles of the sequence, Representing the number of the accumulated data, , The virtual variables representing seasons are used to capture seasonal fluctuations in the time series, such as for a quarterly series, The corresponding parameter vector is ; Then, the parameter vector is estimated by using the least square method The least squares estimation is: wherein: (13) (14) Wherein the matrix In (a) and (b) Is an indication function, i.e To ensure the uniqueness of the least squares solution parameter, The range of the values is as follows ; In the parameter vector And accumulated number The known situation is: 1) The time response function of the dynamic local accumulation gray season model DPDGSM (1, 1) is: (15) Wherein, the Generating a sequence for a dynamic local accumulation A plurality of predicted values; the attenuation coefficient of the model is obtained through least square estimation; generating an initial value of a sequence by dynamic local accumulation at one time; In order to calculate the sum of the variables, ; Is a seasonal virtual variable; 2) The corresponding reduction equation is: (16) Wherein, the Is the first of the original sequence A plurality of predicted values; for the attenuation term, sum term For correcting seasonal variation differences.
  9. 9. The method for predicting the track depth of the asphalt pavement based on sinusoidal driving periodicity according to claim 1, wherein in the step 8, the method is characterized in that the track depth of the asphalt pavement is predicted, and the prediction effect of feet is evaluated by comparing a gray model DPDGSM (1, 1), a time sequence model Holt-windows and a seasonal time sequence model SARIMA-GARCH, a support vector regression SVR model and a machine learning BPNN model through five indexes of APE, MAPES, MAPEP, RMSES and RMSEP.

Description

Asphalt pavement rut local accumulation gray prediction method based on sine drive periodicity Technical Field The invention relates to the technical field of asphalt pavement rut depth prediction, in particular to an asphalt pavement rut local accumulation gray prediction method based on sine driving periodicity. Background The rutting of asphalt pavement is formed due to strain accumulation, so that the pavement deforms along the wheel track, and plastic deformation formed along the wheel track can not only cause discomfort to road users, but also shorten the service life of the pavement. On the other hand, rutting of the road surface may reduce the overall strength and structural stability of the road. These road surface imperfections can significantly impair road surface performance, severely affecting quality of use and service life. Government needs to achieve high quality supplies of pavement structures and full life cycle performance guarantees within established budget constraints, a goal that remains a significant challenge in engineering practice. In this case, the project cost can be reduced and the road surface structure quality can be ensured by using the predictive modeling technology, and in view of the above situation, it is necessary to select a suitable model for predicting the road rut depth. The gray prediction model combines the advantages of a physical model and a data model, a dynamic evolution mechanism after data back is mined through accumulation sum operators in a non-parameterized mode, and a differential equation is used for modeling the accumulation sum sequence, however, a classical gray model usually adopts a global accumulation generation operator (Accumulated Generating Operation, AGO) to preprocess an original sequence, the mode has certain advantages in the aspects of weakening random disturbance and enhancing sequence monotonicity, but the mode endows the same weight to history information, so that early data takes the dominant role in a model modeling process for a long time. When the evolution characteristics of the system change along with time, the global accumulation generating operator easily weakens the action of new information, so that the depicting capability of the model on the recent development trend is reduced. In addition, the model has structural limitation in the face of significant fluctuation and period terms, and the exponential response form of the model is difficult to accurately represent the stepwise change of the pavement index under different time scales, and particularly, when obvious season disturbance or loading stage switching exists, the prediction error can be accumulated and amplified. For the rut depth of the asphalt pavement, the evolution process is influenced by multiple factors such as traffic load increase, environmental temperature change, material performance degradation and the like, and obvious stepwise increase characteristics and non-stationarity are often shown. In the early stage of pavement service, the rut development speed is higher, the increase rate of rut can change along with the gradual stability of the structure, and in the later stage of service, the phenomena of accelerated increase of rut depth can be caused by material aging and structure damage accumulation. Therefore, by adopting a global accumulation mode insensitive to new information, the actual evolution rule of the rut depth is difficult to accurately reflect. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a method for predicting the local accumulated gray of the ruts of the asphalt pavement based on sine driving periodicity, which is superior to the traditional statistical model and the exponential smoothing method, the method also has obvious precision advantage and stability advantage in comparison with a machine learning model, and is more suitable for track depth sequence prediction with obvious trend evolution and seasonal fluctuation characteristics. In order to achieve the above purpose, the present invention adopts the following technical scheme: a method for predicting local accumulated gray of ruts of an asphalt pavement based on sine driving periodicity comprises the following steps: step 1, data acquisition and sequence definition, namely acquiring a target road section rut depth moon detection sequence, and defining an original sequence as ; Step2, introducing a sine driving term and a period correction term to preprocess the data, namely preprocessing the original dataPeriodic processing, introducing sinusoidal driving termsAnd period correction termObtaining a new sequence; Step 3, constructing a seasonal weakening buffer operator SAWBO, wherein the seasonal period is set asWill beSeasonal locationDivided into a plurality of sub-sequences and define the firstThe seasonal average value of each seasonal position isFor the treated season average valueConstructing a seasonal weakening buffer gen